void-cli 0.0.3

//! ls-tree command - list contents of a tree at a given commit.
//!
//! Lists files and directories in a commit's tree, similar to git ls-tree.
//! Supports filtering by path prefix and recursive listing.

use std::collections::BTreeSet;
use std::path::Path;

use serde::Serialize;
use void_core::{
    cid,
    crypto::{CommitReader, EncryptedCommit},

    store::ObjectStoreExt,
};

use crate::context::{open_repo, resolve_ref, void_err_to_cli};
use crate::output::{run_command, CliError, CliOptions};

/// A single entry in the ls-tree output.
#[derive(Debug, Clone, Serialize)]
pub struct TreeEntry {
    /// File mode (040000 for tree, 100644 for blob)
    pub mode: String,
    /// Entry type (tree or blob)
    #[serde(rename = "type")]
    pub entry_type: String,
    /// Full path of the entry
    pub path: String,
    /// Size in bytes (0 for directories)
    pub size: u64,
}

/// JSON output for the ls-tree command.
#[derive(Debug, Clone, Serialize)]
pub struct LsTreeOutput {
    /// CID of the commit
    pub commit: String,
    /// Path prefix filter (if any)
    pub path: String,
    /// List of entries
    pub entries: Vec<TreeEntry>,
}

/// Collect all file entries from the commit's manifest.
fn collect_all_entries(
    store: &impl ObjectStoreExt,
    commit: &void_core::metadata::Commit,
    reader: &CommitReader,
) -> Result<Vec<(String, u64)>, CliError> {
    let manifest = void_core::metadata::manifest_tree::TreeManifest::from_commit(store, commit, reader)
        .map_err(void_err_to_cli)?
        .ok_or_else(|| CliError::internal("commit has no manifest_cid"))?;

    let mut entries: Vec<(String, u64)> = manifest
        .iter()
        .map(|me| {
            let me = me.map_err(void_err_to_cli)?;
            Ok((me.path.clone(), me.length))
        })
        .collect::<Result<_, CliError>>()?;

    entries.sort_by(|a, b| a.0.cmp(&b.0));
    Ok(entries)
}

/// Filter entries to only those under the given path prefix.
fn filter_by_path(entries: Vec<(String, u64)>, path_prefix: &str) -> Vec<(String, u64)> {
    if path_prefix.is_empty() {
        return entries;
    }

    let prefix = if path_prefix.ends_with('/') {
        path_prefix.to_string()
    } else {
        format!("{}/", path_prefix)
    };

    entries
        .into_iter()
        .filter(|(p, _)| p.starts_with(&prefix) || p == path_prefix.trim_end_matches('/'))
        .collect()
}

/// Get the immediate parent directory of a path.
fn parent_dir(path: &str) -> Option<&str> {
    path.rfind('/').map(|idx| &path[..idx])
}

/// Collapse entries to immediate children of the given path prefix (non-recursive mode).
/// Returns both files and inferred directories.
fn collapse_to_immediate_children(
    entries: Vec<(String, u64)>,
    path_prefix: &str,
) -> Vec<TreeEntry> {
    let prefix = if path_prefix.is_empty() {
        String::new()
    } else if path_prefix.ends_with('/') {
        path_prefix.to_string()
    } else {
        format!("{}/", path_prefix)
    };

    // Track directories we've already added
    let mut seen_dirs: BTreeSet<String> = BTreeSet::new();
    let mut result: Vec<TreeEntry> = Vec::new();

    for (path, size) in entries {
        // Skip if path doesn't start with prefix (but handle root case)
        if !prefix.is_empty() && !path.starts_with(&prefix) {
            continue;
        }

        // Get the relative path after the prefix
        let relative = if prefix.is_empty() {
            path.as_str()
        } else {
            &path[prefix.len()..]
        };

        // Check if this is an immediate child or needs directory inference
        if let Some(slash_idx) = relative.find('/') {
            // This file is in a subdirectory
            let dir_name = &relative[..slash_idx];
            let full_dir_path = if prefix.is_empty() {
                format!("{}/", dir_name)
            } else {
                format!("{}{}/", prefix, dir_name)
            };

            if !seen_dirs.contains(&full_dir_path) {
                seen_dirs.insert(full_dir_path.clone());
                result.push(TreeEntry {
                    mode: "040000".to_string(),
                    entry_type: "tree".to_string(),
                    path: full_dir_path,
                    size: 0,
                });
            }
        } else {
            // This is an immediate child file
            result.push(TreeEntry {
                mode: "100644".to_string(),
                entry_type: "blob".to_string(),
                path,
                size,
            });
        }
    }

    // Sort by path
    result.sort_by(|a, b| a.path.cmp(&b.path));
    result
}

/// Expand all entries (recursive mode) - include files and their parent directories.
fn expand_recursive(entries: Vec<(String, u64)>, path_prefix: &str) -> Vec<TreeEntry> {
    let prefix = if path_prefix.is_empty() {
        String::new()
    } else if path_prefix.ends_with('/') {
        path_prefix.to_string()
    } else {
        format!("{}/", path_prefix)
    };

    // Track directories we've seen
    let mut seen_dirs: BTreeSet<String> = BTreeSet::new();
    let mut result: Vec<TreeEntry> = Vec::new();

    for (path, size) in entries {
        // Skip if path doesn't start with prefix (but handle root case)
        if !prefix.is_empty() && !path.starts_with(&prefix) {
            continue;
        }

        // Add all parent directories
        let mut current = path.as_str();
        while let Some(parent) = parent_dir(current) {
            if !parent.is_empty() {
                let dir_path = format!("{}/", parent);
                // Only add if within the path prefix
                if prefix.is_empty()
                    || dir_path.starts_with(&prefix)
                    || prefix.starts_with(&dir_path)
                {
                    if !seen_dirs.contains(&dir_path) {
                        seen_dirs.insert(dir_path.clone());
                    }
                }
            }
            current = parent;
        }

        // Add the file entry
        result.push(TreeEntry {
            mode: "100644".to_string(),
            entry_type: "blob".to_string(),
            path,
            size,
        });
    }

    // Add directory entries
    for dir in seen_dirs {
        // Only include directories that are within or descendants of the prefix
        if prefix.is_empty() || dir.starts_with(&prefix) {
            result.push(TreeEntry {
                mode: "040000".to_string(),
                entry_type: "tree".to_string(),
                path: dir,
                size: 0,
            });
        }
    }

    // Sort by path
    result.sort_by(|a, b| a.path.cmp(&b.path));
    result
}

/// Run the ls-tree command.
///
/// Lists files and directories in a commit's tree.
///
/// # Arguments
/// * `cwd` - Current working directory
/// * `commit_ref` - Commit reference (HEAD, branch, tag, or CID)
/// * `path_filter` - Optional path prefix to filter
/// * `name_only` - Show only filenames (no mode/type/size)
/// * `recursive` - Recurse into subdirectories
/// * `opts` - CLI options
pub fn run(
    cwd: &Path,
    commit_ref: &str,
    path_filter: Option<&str>,
    name_only: bool,
    recursive: bool,
    opts: &CliOptions,
) -> Result<(), CliError> {
    run_command("ls-tree", opts, |ctx| {
        ctx.progress("Loading tree...");
        ctx.verbose("Reading repository context...");

        let repo = open_repo(cwd)?;

        ctx.verbose(format!("Resolving ref: {}", commit_ref));

        // Resolve the commit reference
        let commit_cid_typed = resolve_ref(repo.void_dir().as_std_path(), commit_ref)?;
        let commit_cid = cid::from_bytes(commit_cid_typed.as_bytes())
            .map_err(|e| CliError::internal(format!("invalid commit CID: {e}")))?;
        let commit_cid_str = commit_cid.to_string();

        ctx.verbose(format!(
            "Commit: {}",
            &commit_cid_str[..12.min(commit_cid_str.len())]
        ));

        // Create object store
        let store = repo.store().map_err(void_err_to_cli)?;

        // Read commit using CommitReader (handles VD01 format)
        let commit_encrypted: EncryptedCommit = store
            .get_blob(&commit_cid)
            .map_err(|e| CliError::not_found(format!("commit not found: {e}")))?;
        let (commit_bytes, reader) = CommitReader::open_with_vault(repo.vault(), &commit_encrypted)
            .map_err(|e| CliError::internal(format!("commit decryption failed: {e}")))?;
        let commit = commit_bytes.parse()
            .map_err(|e| CliError::internal(format!("failed to parse commit: {e}")))?;

        ctx.verbose("Collecting file entries...");

        let all_entries = collect_all_entries(&store, &commit, &reader)?;

        let path_prefix = path_filter.unwrap_or("");

        // Filter by path if specified
        let filtered_entries = filter_by_path(all_entries, path_prefix);

        // Build tree entries based on recursive flag
        let tree_entries = if recursive {
            expand_recursive(filtered_entries, path_prefix)
        } else {
            collapse_to_immediate_children(filtered_entries, path_prefix)
        };

        ctx.progress(format!("Found {} entries", tree_entries.len()));

        // Format human-readable output
        if !ctx.use_json() {
            for entry in &tree_entries {
                if name_only {
                    ctx.info(&entry.path);
                } else {
                    // git-like format: mode type cid\tpath
                    // Since we don't have a separate CID for each file, we show "-"
                    ctx.info(format!(
                        "{} {} -\t{}",
                        entry.mode, entry.entry_type, entry.path
                    ));
                }
            }
        }

        Ok(LsTreeOutput {
            commit: commit_cid_str,
            path: path_prefix.to_string(),
            entries: tree_entries,
        })
    })
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_parent_dir() {
        assert_eq!(parent_dir("file.txt"), None);
        assert_eq!(parent_dir("src/main.rs"), Some("src"));
        assert_eq!(parent_dir("a/b/c.txt"), Some("a/b"));
    }

    #[test]
    fn test_filter_by_path_empty() {
        let entries = vec![
            ("src/main.rs".to_string(), 100),
            ("README.md".to_string(), 50),
        ];
        let filtered = filter_by_path(entries.clone(), "");
        assert_eq!(filtered.len(), 2);
    }

    #[test]
    fn test_filter_by_path_prefix() {
        let entries = vec![
            ("src/main.rs".to_string(), 100),
            ("src/lib.rs".to_string(), 80),
            ("README.md".to_string(), 50),
        ];
        let filtered = filter_by_path(entries, "src");
        assert_eq!(filtered.len(), 2);
        assert!(filtered.iter().all(|(p, _)| p.starts_with("src/")));
    }

    #[test]
    fn test_collapse_to_immediate_children_root() {
        let entries = vec![
            ("README.md".to_string(), 50),
            ("src/main.rs".to_string(), 100),
            ("src/lib.rs".to_string(), 80),
            ("tests/test.rs".to_string(), 60),
        ];

        let result = collapse_to_immediate_children(entries, "");

        // Should have: README.md (blob), src/ (tree), tests/ (tree)
        assert_eq!(result.len(), 3);

        let readme = result.iter().find(|e| e.path == "README.md").unwrap();
        assert_eq!(readme.entry_type, "blob");

        let src = result.iter().find(|e| e.path == "src/").unwrap();
        assert_eq!(src.entry_type, "tree");
    }

    #[test]
    fn test_collapse_to_immediate_children_subdir() {
        let entries = vec![
            ("src/main.rs".to_string(), 100),
            ("src/lib.rs".to_string(), 80),
            ("src/utils/helper.rs".to_string(), 40),
        ];

        let result = collapse_to_immediate_children(entries, "src");

        // Should have: main.rs, lib.rs (blobs), utils/ (tree)
        assert_eq!(result.len(), 3);

        let main_rs = result.iter().find(|e| e.path == "src/main.rs").unwrap();
        assert_eq!(main_rs.entry_type, "blob");

        let utils = result.iter().find(|e| e.path == "src/utils/").unwrap();
        assert_eq!(utils.entry_type, "tree");
    }

    #[test]
    fn test_expand_recursive() {
        let entries = vec![
            ("src/main.rs".to_string(), 100),
            ("src/utils/helper.rs".to_string(), 40),
        ];

        let result = expand_recursive(entries, "");

        // Should include: src/ (tree), src/utils/ (tree), src/main.rs (blob), src/utils/helper.rs (blob)
        assert_eq!(result.len(), 4);

        let trees: Vec<_> = result.iter().filter(|e| e.entry_type == "tree").collect();
        let blobs: Vec<_> = result.iter().filter(|e| e.entry_type == "blob").collect();

        assert_eq!(trees.len(), 2);
        assert_eq!(blobs.len(), 2);
    }

    #[test]
    fn test_tree_entry_serialization() {
        let entry = TreeEntry {
            mode: "100644".to_string(),
            entry_type: "blob".to_string(),
            path: "src/main.rs".to_string(),
            size: 1234,
        };

        let json = serde_json::to_string(&entry).unwrap();
        assert!(json.contains("\"mode\":\"100644\""));
        assert!(json.contains("\"type\":\"blob\""));
        assert!(json.contains("\"path\":\"src/main.rs\""));
        assert!(json.contains("\"size\":1234"));
    }

    #[test]
    fn test_ls_tree_output_serialization() {
        let output = LsTreeOutput {
            commit: "bafytest123".to_string(),
            path: "src/".to_string(),
            entries: vec![TreeEntry {
                mode: "100644".to_string(),
                entry_type: "blob".to_string(),
                path: "src/main.rs".to_string(),
                size: 100,
            }],
        };

        let json = serde_json::to_string(&output).unwrap();
        assert!(json.contains("\"commit\":\"bafytest123\""));
        assert!(json.contains("\"path\":\"src/\""));
        assert!(json.contains("\"entries\""));
    }
}